The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
ESSAYS ON ISSUES THE FEDERAL RESERVE BANK OF CHICAGO MAY 2003 NUMBER 189 Chicago Fed Letter Regional growth in worker quality by Daniel Aaronson, economic advisor, and Daniel Sullivan, vice president An analysis of regional labor quality growth places the Midwest near the center of the U.S. labor quality distribution. To maintain its relative position going forward, the Midwest will have to increase its share of workers with post-secondary education. This will require policies that not only promote the education of its residents, but also efforts to make the region more attractive to the highly mobile population of college graduates. The rate of improvement in work force quality is among the key determinants of an economy’s potential rate of growth. Indeed, in previous work, we have shown that increases in a simple labor quality measure that quantifies the effects of such factors as educational attainment and labor market experience ac1. Proportion of workers with at least high school count for roughly 10% of U.S. labor productivproportion 0.96 ity growth since 1964.1 Midwest Moreover, this contribu0.86 tion of worker quality High region U.S. improvement to eco0.76 nomic growth has varied substantially over 0.66 time, reflecting variaLow region tion in the rate of in0.56 crease in educational attainment and the 0.46 aging of the work force. However, over this peri0.36 1964 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’96 ’00 od, not all regions of the U.S. developed human SOURCE: Authors’ calculations based on data from the US. Census Bureau and Bureau of Labor Statistics, 1964–2002, Current Population Survey, capital at the same rate. March supplements. In this Chicago Fed Letter, we provide estimates of the rate of improvement in the quality of regional work forces, with particular emphasis on the Midwest. Although we show that broad patterns of human capital accumulation are similar across regions, there are some notable contrasts as well. In particular, regions, chiefly in the South, that historically have had relatively low levels of human capital have been slowly converging to the levels of regions, particularly the West, that historically have had relatively high levels of human capital. This convergence has been driven primarily by a reduction in disparities in educational attainment, particularly high school graduation. By our measure, average worker quality in the Midwest is strikingly similar to the overall U.S. average. In fact, this similarity has existed, without interruption, since the 1960s. Though the Midwest is close to the national average in overall human capital, it arrives at that average in a particularly midwestern manner, with a strong showing at the lower end of the educational distribution and a less impressive performance at the high end. As a result, in the Midwest, educational attainment is concentrated more in the middle ranges, with both fewer high school dropouts and fewer college graduates than in the nation. This, perhaps, poses a challenge for the region. Because a high school education is already nearly universal in the Midwest, further reductions in the fraction of dropouts may be more difficult to achieve. To keep up with the nation in the development in overall human capital, the Midwest with obtaining a college century. During this period, high school degree likely depends graduation went from a rarity to the on the program of norm. College attendance and graduaproportion 0.70 study, the quality of the tion rates also rose rapidly during the institution, and a myritwentieth century, especially after WWII 0.60 ad of other factors. But with the introduction of the GI Bill and High region available data sources increased federal funding of higher ed0.50 merely record whether ucation. Increasing graduation rates U.S. a worker has a college have led to a corresponding increase 0.40 degree. Similarly, the in the percentage of workers with high Low region productivity increase school and college education. Over the 0.30 associated with a year last several decades, the increase in edMidwest of work experience will ucational attainment has generally been 0.20 vary with the nature of greater in regions that started at lower the work, how much levels, leading to some convergence in 0.10 1964 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’96 ’00 time is devoted to train- regional educational attainment. ing, and other factors. SOURCE: Authors’ calculations based on data from the US. Census Bureau Figure 1 shows the share of workers and Bureau of Labor Statistics, 1964–2002, Current Population Survey, But data sources do not March supplements. with a high school diploma in the U.S. include such informaand the East North Central Census retion. Indeed, our meagion, which we will refer to as the Midmay need to improve its relative persure of worker quality growth is based west. In addition, the figure shows the formance in training, attracting, and on proxies for years of work experience levels of high school attainment in the retaining college graduates. derived from the difference between a Census regions with the highest and worker’s age and their years of formal lowest fraction of high-school-educated What is worker quality? education. Unobserved differences in 2 Our measure of worker quality is derived time out of the labor force due, for ex- workers. In 1964, 57.7% of U.S. workers had completed high school, with from economic models of human cap- ample, to raising children, will lead such the rate varying from 44.0% in the East ital accumulation. Workers are assumed proxies to differ from actual experito invest in productivity-increasing skills ence. Though imperfect, our measure South Central region to 68.0% in the West region. By 2001, the last year for through formal education and on-the- of labor quality does, however, sumwhich data are available, the overall job training, while firms are assumed marize the impact of observable labor rate had increased to 89.4%, varying to hire additional labor until workers’ force characteristics on productivity. from 85.2% in the West South Central marginal productivity coincides with region to 92.5% in the West North Trends in human capital their wage rate. The latter assumption accumulation Central region. In general, the progress allows us to infer the effects of worker was more rapid in the regions that We begin by documenting some of the characteristics on productivity, which started at low levels of high school broad trends in educational attainare not directly observable, from their ment that underlie our effects on predicted wages, which can estimates of worker be estimated from cross-sectional data 3. Proportion of workers with college degree on workers’ wage rates and indicators of quality growth and show how these trends differ proportion their human capital. In particular, we 0.35 across regions. We focus use data on individual workers from on educational levels the Current Population Survey (CPS) 0.30 because, while factors to estimate statistical models of wage High region U.S. such as the aging of the determination that value workers’ ed0.25 work force and the inucation, experience, and other characcrease in female labor teristics. We then apply these value 0.20 force participation have Midwest estimates to the changing distributions significant impacts on Low region of worker characteristics to obtain esti0.15 the growth of labor mates of the growth in overall worker quality over time, they quality. 0.10 have less impact in difTo be sure, the available data on worker ferentiating labor quali0.05 characteristics only begin to scratch the ty across regions. 1964 ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’96 ’00 surface in explaining the determinants SOURCE: Authors’ calculations based on data from the US. Census Bureau and Bureau of Labor Statistics, 1964–2002, Current Population Survey, of wages and productivity. For instance, U.S. levels of formal edMarch supplements. ucation have expanded the productivity increase associated greatly over the last 2. Proportion of workers with at least some college 4. Labor quality 1980$ 5.50 5.25 U.S. 5.00 High region Low region 4.75 Midwest 4.50 4.25 1964 In 1964, rates of postsecondary attainment ranged from 16.9% in the East South Central region to 31.3% in the West region, a range of 15.4 percentage points. In 2001, the same two regions were at the top and bottom, but their rates had risen to 53.2% and 61.9%, respectively, a gap of only 8.7 percentage points. In contrast to its relatively high levels of high SOURCE: Authors’ calculations based on data from the US. Census Bureau school attainment, until and Bureau of Labor Statistics, 1964–2002, Current Population Survey, March supplements. very recently the Midwest had lagged behind the nation in its rate of attainment, leading to the convergence post-secondary educational attainment. From the mid-1960s through the late evident in the figure. 1980s, workers in the Midwest were ofSince the mid-1960s, Midwest rates of ten two percentage points or more less high school attainment have nearly allikely to have had some college educaways exceeded those in the rest of the tion than the typical U.S. worker. Hownation. From the late 1960s through the ever, that gap has narrowed since the early 1980s, the gap was typically quite early 1990s, and the Midwest rate was small. Since the mid-1980s, however, only a tenth of a percentage point bethe Midwest has seen noticeably more low the national average in 2001. Given progress in high school attainment than that increases in rates of post-secondthe nation as a whole. Indeed in 1998 ary educational attainment show no and 1999, the Midwest recorded the sign of slowing, continued progress in highest regional rates of high school raising such rates is likely to become attainment, and its 2001 level of 91.8% increasingly important to maintaining was less than a percentage point below the Midwest’s relative position in overthe leading region. Given the very low all labor quality. relative wages earned by workers without a high school diploma, the Midwest’s Figure 3 shows that, for the country as a relatively strong showing in high school whole, the fraction of college graduates attainment has been a significant factor increased from 11.8% in 1964 to 28.4% in 2001. Unlike the convergence we have in boosting average labor quality in the seen in regional rates of high school and region. However, rates of increase have post-secondary educational attainment, slowed, as high school attainment has regional rates of college completion have become nearly universal, limiting the actually diverged. In 1964, college gradboost to labor quality that can come uation rates ranged from 9.2% in the from this avenue in the future. East South Central region to 15.1% in Figure 2 displays the growth in the fracthe West region, a spread of 5.9 percenttion of workers with at least some postage points, while in 2001 the range was 3 secondary education. For the nation as from 23.1% in the East South Central a whole, the fraction of such workers inregion to 32.8 % in the New England creased from 22.6% in 1964 to 58.0% region, a spread of 8.9 percentage points. in 2001. Regional rates of post-secondary educational attainment also converged, Figure 3 also shows that the Midwest has long lagged behind the rest of the nathough somewhat less dramatically than tion in the fraction of workers with a in the case of high school attainment. ’68 ’72 ’76 ’80 ’84 ’88 ’92 ’96 ’00 college degree. Indeed, the figure suggests that the gap relative to the rest of the nation may have increased slightly after about 1980. The Midwest rate of 27.1% in 2001 was 1.3 percentage points below the national average. Because workers with college degrees tend to be more mobile than other workers, factors other than the generosity of support for higher education influence the share of a region’s work force with degrees. The Midwest may be at a disadvantage in attracting and retaining college-educated workers. Still, given its current position, increasing the rate of college graduates in its work force represents an opportunity for the Midwest to improve its labor quality. The Worker Quality Index Our composite estimates of labor quality summarize the impact on productivity of the above trends in educational attainment, along with other changes in the work force. To derive these estimates, we use the 1980 CPS data for the nation as a whole to estimate a statistical model that relates wage rates to workers’ education, potential experience, sex, and other characteristics. We then use that model to predict what workers in other years would have earned in 1980. The average of these predictions is worker quality. By focusing on the growth of such predicted wages, we isolate the portion of wage growth that is due to the changing composition of the work force Michael H. Moskow, President; William C. Hunter, Senior Vice President and Director of Research; Douglas Evanoff, Vice President, financial studies; David Marshall, team leader, macroeconomic policy research; Daniel Sullivan, Vice President, microeconomic policy research; William Testa, Vice President, regional programs and Economics Editor; Helen O’D. Koshy, Editor; Kathryn Moran, Associate Editor. Chicago Fed Letter is published monthly by the Research Department of the Federal Reserve Bank of Chicago. The views expressed are the authors’ and are not necessarily those of the Federal Reserve Bank of Chicago or the Federal Reserve System. Articles may be reprinted if the source is credited and the Research Department is provided with copies of the reprints. Chicago Fed Letter is available without charge from the Public Information Center, Federal Reserve Bank of Chicago, P.O. Box 834, Chicago, Illinois 60690-0834, tel. 312-322-5111 or fax 312-322-5515. Chicago Fed Letter and other Bank publications are available on the World Wide Web at http:// www.chicagofed.org. ISSN 0895-0164 rather than accumulation of physical capital or technical progress. These latter factors raise the value of an hour supplied by a worker with a constant level of human capital. Figure 4 displays our estimates for 1964 through 2001 for the U.S., the Midwest, and the regions with the lowest and highest labor quality. The typical worker of 1964 would have earned about $4.62 per hour in the labor market of 1980, while the typical worker of 2001 would have earned about $5.16 in 1980. This represents an 11.7% increase in labor quality or about three-tenths of a percentage point increase per year. Though changes in the distribution of potential experience and the fraction of women in the labor force had important effects on the rate of labor quality growth over certain periods of time, most of the overall increase since the mid-1960s is due to the increases in educational attainment documented above. It is also clear from figure 4 that labor quality in the various regions of the country has converged to a significant extent since the mid-1960s. For instance, in 1964, labor quality in the West (the region with highest labor quality) was more than 10% higher than in the southern regions. By 2001, however, the gap had shrunk to about 5%. This was largely due to the faster increase in the fraction of workers with high school or some college education that occurred in the states that started with lower labor quality. Throughout the period of our data, overall Midwest labor quality has been extremely close to the national average. Indeed, though not shown in the figure, each of the five states in the Midwest region are tightly bunched around this norm, with Michigan just slightly above and Indiana and Wisconsin just slightly below the national average. As shown above, this reflects the offsetting effects of having both fewer high school dropouts and fewer college graduates. poses a problem going for ward. With the share of workers with high school degrees already quite high, it will be difficult to increase it significantly further. To keep up with the rest of the nation, the Midwest will have to increase its share of workers with post-secondary education. This will require policies that not only promote the education of its residents, but also those that make the region attractive to the highly mobile population of college graduates. 1 See Daniel Aaronson and Daniel Sullivan, 2001, “Growth in worker quality,” Economic Perspectives, Vol. 25, No. 4, pp. 53–74. This paper also contains a detailed discussion of our methodology. 2 Data come from the annual March supplements to the CPS. The East North Central region is composed of Ohio, Michigan, Indiana, Illinois, and Wisconsin. There are eight other Census regions. 3 A portion of the relatively large increase in the fraction of workers with some college education that occurred in 1992 is due to a change in the CPS educational attainment question. Aaronson and Sullivan (2001) discuss how we mitigate the effects of this change on our measure of labor quality. 4 In the standard growth accounting framework, the contribution of labor quality to labor productivity is the product of the growth in labor quality and the share of labor in total costs, which is about two-thirds. Conclusion Labor quality has risen significantly since the mid-1960s. The 11.7% increase in our index of labor quality accounts for about 10% of the growth in U.S. labor productivity over the period.4 Labor quality has, moreover, converged across the various regions of the country, with the gap between the highest and lowest regions now about half as large in percentage terms as it was in the mid1960s. The Midwest has been, and continues to be, very near the national average of labor quality. The fact that the Midwest has gotten to its current position near the center of the labor quality distribution by having a higher than average share of workers with a high school education perhaps